Users may issue queries to a virtual assistant, search engine, or other technological interface to seek help when experiencing acute emotional distress. Current approaches surface relevant information, such as the number for a local helpline, that can help connect the users to resources to deal with the matter. Currently, such information is surfaced if a spoken query to a voice-based virtual assistant contains specific trigger words related to mental health and emotional well-being. Such a simplistic approach lacks a nuanced understanding of the user’s context and state of mind when issuing the query. This disclosure describes techniques to infer the user’s state of mind when issuing a query to a voice-based virtual assistant. The user’s state of mind inferred with permission in combination with the content of the query is used to surface supportive resources most relevant to the query. A model can be trained to infer emotional states of mind across a diversity of cultures, languages, genders, ethnicities, etc.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.